Visualization of the Sleep Data for Sleep Cluster Nos 2 and 5¶
In [23]:
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import math
import os
%matplotlib inline
In [3]:
data_series = pd.read_csv("Series1_2525_ClusterAssign.csv", sep=',', header = None)
In [4]:
#30 seconds - 2 waves each
x1 = data_series.loc[0:3750,0]
y1 = data_series.loc[0:3750,1]
z1 = data_series.loc[0:3750,2]
c1 = data_series.loc[0:3750,3]
n1 = np.size(c1)
t1 = [int(i) for i in range(0,n1)]
start_p = 0
end_p = 3751
x2 = data_series.loc[243751:247501,0]
y2 = data_series.loc[243751:247501,1]
z2 = data_series.loc[243751:247501,2]
c2 = data_series.loc[243751:247501,3]
n2 = np.size(x2)
t2 = [int(i+15000) for i in range(0,n2)]
fig = plt.figure(figsize=(30,10))
ax1 = fig.add_subplot(311)
ax2 = fig.add_subplot(312)
ax3 = fig.add_subplot(313)
t= []
x = x1
x = x.append(x2)
y = y1
y = y.append(y2)
z = z1
z = z.append(x2)
c = c1
c = c.append(c2)
#print(c)
d = np.column_stack((x,y,z))
dc = np.column_stack((x,y,z,c))
d1 = pd.DataFrame(d)
dc = pd.DataFrame(dc)
#print(d)
start_p1 = 0
end_p1 = 3750
start_p2 = 3750
end_p2 = 7500
ax1.plot(d1.loc[:,0],'k')
ax2.plot(d1.loc[:,1],'k')
ax3.plot(d1.loc[:,2],'k')
p = ax1.axvspan(start_p1, end_p1, facecolor='g', alpha=0.5)
q = ax1.axvspan(start_p2, end_p2, facecolor='r', alpha=0.5)
p = ax2.axvspan(start_p1, end_p1, facecolor='g', alpha=0.5)
q = ax2.axvspan(start_p2, end_p2, facecolor='r', alpha=0.5)
p = ax3.axvspan(start_p1, end_p1, facecolor='g', alpha=0.5)
q = ax3.axvspan(start_p2, end_p2, facecolor='r', alpha=0.5)
In [5]:
d = pd.DataFrame(d)
dc = pd.DataFrame(dc)
d = d.append(d)
d.to_csv("Series1_30_Repv1.csv", sep =',', header = False, index = False)
dc = dc.append(dc)
dc.to_csv("Series1_30_Rep_Label.csv", sep =',', header = False, index = False)
In [3]:
data_TICC = pd.read_csv("Series1_30_Repv1-Cluster-5200.csv", sep =',',header = None)
print(data_TICC)
In [8]:
fig = plt.figure(figsize=(50,10),dpi = 400)
ax1 = fig.add_subplot(211)
ax2 = fig.add_subplot(212)
#fig.tight_layout()
plt.subplots_adjust(hspace=0.2)
x1 = data_TICC.loc[:,0]
n = len(x1)
print(n)
clust = data_TICC.loc[:,3]
clust_num = clust.unique()
print(clust_num)
#x = [int(i) for i in x]
#print(str(howmanyclust) + " <-- this many clusters")
#print(clust)
clust = np.asarray(clust, dtype = int)
#y = column(data, 1)
c = ['b', 'g', 'r', 'c', 'm', 'y']
ax1.plot(x1,'k')
#print(np.shape(clust))
mini = 0
currcol = nxtcol = 0
for i in range(1,n):
currcol = c[clust[i-1]]
nxtcol = c[clust[i]]
if((i+1)==n):
p = ax2.axvspan(mini, i-1, facecolor=c[clust[i]], alpha=0.5)
if nxtcol == currcol:
if mini == 0:
mini = i-1
continue
else:
j = clust[i-1]
p = ax2.axvspan(mini, i-1, facecolor=c[j], alpha=0.5)
mini = 0
ax2.set_title('Clustering by Experts', fontsize = 32)
print(mini, i)
#Expert Label:
data_label = pd.read_csv("Series1_30_Rep_Label.csv", sep =',',header=None)
#print(data_label)
x2 = data_label.loc[:,0]
n = len(x2)
print(n)
clust2 = data_label.loc[:,3]
clust_num2 = clust2.unique()
clust2 = np.asarray(clust2, dtype = int)
print(clust2)
ax2.plot(x2,'k')
#print(np.shape(clust2))
mini = 0
i=0
currcol = nxtcol = 0
for i in range(1,n):
currcol = c[clust2[i-1]]
nxtcol = c[clust2[i]]
if((i+1)==n):
p = ax1.axvspan(mini, i-1, facecolor=c[clust2[i]], alpha=0.5)
if nxtcol == currcol:
if mini == 0:
mini = i-1
continue
else:
j = clust2[i-1]
p = ax1.axvspan(mini, i-1, facecolor=c[j], alpha=0.5)
mini = 0
ax1.set_title('Clustering by TICC', fontsize = 32)
#fig.savefig("TICC vs Expert")
In [2]:
fig = plt.figure(figsize=(50,10),dpi = 400)
x1 = data_TICC.loc[:,0]
n = len(x1)
print(n)
clust = data_TICC.loc[:,3]
clust_num = clust.unique()
print(clust_num)
clust = np.asarray(clust, dtype = int)
#y = column(data, 1)
c = ['b', 'g', 'r', 'c', 'm', 'y']
plt.plot(x1,'k')
#print(np.shape(clust))
mini = 0
currcol = nxtcol = 0
for i in range(1,n):
currcol = c[clust[i-1]]
nxtcol = c[clust[i]]
if((i+1)==n):
p = plt.axvspan(mini, i-1, facecolor=c[clust[i]], alpha=0.5)
if nxtcol == currcol:
if mini == 0:
mini = i-1
continue
else:
j = clust[i-1]
p = plt.axvspan(mini, i-1, facecolor=c[j], alpha=0.5)
mini = 0
#plt.title('Clustering by TICC', fontsize = 64)
print(mini, i)
#fig.savefig("Clustering by TICC")
In [5]:
c = ['b', 'g', 'r', 'c', 'm', 'y']
In [7]:
#Expert Label:
fig = plt.figure(figsize=(50,10),dpi = 400)
data_label = pd.read_csv("Series1_30_Rep_Label.csv", sep =',',header=None)
#print(data_label)
x2 = data_label.loc[:,0]
n = len(x2)
print(n)
clust2 = data_label.loc[:,3]
clust_num2 = clust2.unique()
clust2 = np.asarray(clust2, dtype = int)
print(clust2)
plt.plot(x2,'k')
#print(np.shape(clust2))
mini = 0
i=0
currcol = nxtcol = 0
for i in range(1,n):
currcol = c[clust2[i-1]]
nxtcol = c[clust2[i]]
if((i+1)==n):
p = plt.axvspan(mini, i-1, facecolor=c[clust2[i]], alpha=0.5)
if nxtcol == currcol:
if mini == 0:
mini = i-1
continue
else:
j = clust2[i-1]
p = plt.axvspan(mini, i-1, facecolor=c[j], alpha=0.5)
mini = 0
#plt.title('Clustering by Expert', fontsize = 64)
#fig.savefig("Clustering by Expert")
In [31]:
#TICC Clustering for all the sensors
fig = plt.figure(figsize=(20,20),dpi=400)
ax1 = fig.add_subplot(3,1,1)
ax2 = fig.add_subplot(3,1,2)
ax3 = fig.add_subplot(313)
x1 = data_TICC.loc[:,0]
x2 = data_TICC.loc[:,1]
x3 = data_TICC.loc[:,2]
n = len(x1)
print(n)
clust = data_TICC.loc[:,3]
clust_num = clust.unique()
#print(clust)
clust = np.asarray(clust, dtype = int)
#y = column(data, 1)
c = ['b', 'g', 'r', 'c', 'm', 'y']
ax1.plot(x1,'k')
ax2.plot(x2,'k')
ax3.plot(x3,'k')
#print(np.shape(clust))
mini = 0
currcol = nxtcol = 0
for i in range(1,n):
currcol = c[clust[i-1]]
nxtcol = c[clust[i]]
if((i+1)==n):
p1 = ax1.axvspan(mini, i-1, facecolor=c[clust[i]], alpha=0.5)
p2 = ax2.axvspan(mini, i-1, facecolor=c[clust[i]], alpha=0.5)
p3 = ax3.axvspan(mini, i-1, facecolor=c[clust[i]], alpha=0.5)
if nxtcol == currcol:
if mini == 0:
mini = i-1
continue
else:
j = clust[i-1]
p1 = ax1.axvspan(mini, i-1, facecolor=c[clust[i-1]], alpha=0.5)
p2 = ax2.axvspan(mini, i-1, facecolor=c[clust[i-1]], alpha=0.5)
p3 = ax3.axvspan(mini, i-1, facecolor=c[clust[i-1]], alpha=0.5)
#plt.title('Clustering by TICC', fontsize = 64)
mini = 0
print(mini, i)
fig.savefig("Clustering by TICC+A.")
In [33]:
#Expert Label:
fig = plt.figure(figsize=(20,20),dpi = 400)
data_label = pd.read_csv("Series1_30_Rep_Label.csv", sep =',',header=None)
#print(data_label)
y1 = data_label.loc[:,0]
y2 = data_label.loc[:,1]
y3 = data_label.loc[:,2]
ax1 = fig.add_subplot(3,1,1)
ax2 = fig.add_subplot(3,1,2)
ax3 = fig.add_subplot(313)
n = len(x2)
print(n)
clust2 = data_label.loc[:,3]
clust_num2 = clust2.unique()
clust2 = np.asarray(clust2, dtype = int)
print(clust2)
ax1.plot(y1,'k')
ax2.plot(y2,'k')
ax3.plot(y3,'k')
#print(np.shape(clust2))
mini = 0
i=0
currcol = nxtcol = 0
for i in range(1,n):
currcol = c[clust2[i-1]]
nxtcol = c[clust2[i]]
if((i+1)==n):
p1 = ax1.axvspan(mini, i-1, facecolor=c[clust2[i]], alpha=0.5)
p2 = ax2.axvspan(mini, i-1, facecolor=c[clust2[i]], alpha=0.5)
p3 = ax3.axvspan(mini, i-1, facecolor=c[clust2[i]], alpha=0.5)
if nxtcol == currcol:
if mini == 0:
mini = i-1
continue
else:
j = clust2[i-1]
p1 = ax1.axvspan(mini, i-1, facecolor=c[clust2[i-1]], alpha=0.5)
p2 = ax2.axvspan(mini, i-1, facecolor=c[clust2[i-1]], alpha=0.5)
p3 = ax3.axvspan(mini, i-1, facecolor=c[clust2[i-1]], alpha=0.5)
mini = 0
#plt.title('Clustering by Expert', fontsize = 64)
fig.savefig("Clustering by Expert - All Sensors!")
In [ ]: